mersenneforum.org  

Go Back   mersenneforum.org > Search Forums

Showing results 1 to 25 of 467
Search took 0.05 seconds.
Search: Posts Made By: charybdis
Forum: CADO-NFS 2021-09-24, 16:09
Replies: 0
Views: 67
Posted By charybdis
AVX-512 support in latest CADO

The latest revision to CADO includes improvements to the lattice reduction routine, including the addition of AVX-512 code for what I believe is the first time. There is a speedup of a few percent in...
Forum: Msieve 2021-09-24, 15:50
Replies: 60
Sticky: Msieve GPU LA
Views: 3,049
Posted By charybdis
If lims have to stay at 250M, it would probably...

If lims have to stay at 250M, it would probably be possible to stretch the upper limit of doable jobs a bit by using 3LP on both sides to catch some of the relations that are lost due to the low...
Forum: Msieve 2021-09-24, 12:36
Replies: 60
Sticky: Msieve GPU LA
Views: 3,049
Posted By charybdis
Also 250M is very low for alim/rlim at this size;...

Also 250M is very low for alim/rlim at this size; some quick testing suggests the optimum is likely between 500M and 1000M. Is this done to keep memory use low? How many 16f contributors don't have...
Forum: Msieve 2021-09-23, 12:46
Replies: 60
Sticky: Msieve GPU LA
Views: 3,049
Posted By charybdis
@frmky, for future reference, when I tested this...

@frmky, for future reference, when I tested this I found that rational side sieving with *algebraic* 3LP was fastest. This shouldn't be too much of a surprise: the rational norms are larger, but not...
Forum: Aliquot Sequences 2021-09-21, 00:19
Replies: 3,115
Views: 249,749
Posted By charybdis
What Q value have you sieved up to, and what is...

What Q value have you sieved up to, and what is the overall duplication rate like? I'd be interested to know how well my estimates from test-sieving have held up.
Forum: Cunningham Tables 2021-09-19, 16:22
Replies: 76
Views: 30,763
Posted By charybdis
Wow! This beats the polynomial used for RSA-200...

Wow! This beats the polynomial used for RSA-200 in 2005! (3.464e-15) :w00t:
Forum: Msieve 2021-09-19, 00:30
Replies: 60
Sticky: Msieve GPU LA
Views: 3,049
Posted By charybdis
I tested the binary from here...

I tested the binary from here (https://www.mersenneforum.org/showpost.php?p=470249&postcount=10) on 2,2174L with 34-bit large primes and it seemed to work fine. Yield was more than double that at...
Forum: Msieve 2021-09-18, 14:37
Replies: 60
Sticky: Msieve GPU LA
Views: 3,049
Posted By charybdis
Or is this just the limit for 16e with 33-bit...

Or is this just the limit for 16e with 33-bit large primes? I know you've avoided going higher because of the difficulty of the LA and the msieve filtering bug, but now that the bug is fixed and GPUs...
Forum: Cunningham Tables 2021-09-12, 20:44
Replies: 76
Views: 30,763
Posted By charybdis
I haven't run anything beyond basic CADO poly...

I haven't run anything beyond basic CADO poly selection. Feel free to try spinning with EdH's script. I think Max does some more things that are known only to him.
Forum: Cunningham Tables 2021-09-12, 15:30
Replies: 76
Views: 30,763
Posted By charybdis
Continued up to 35M. Best score was 2.807e-15,...

Continued up to 35M. Best score was 2.807e-15, probably good enough for sieving?

n:...
Forum: NFS@Home 2021-09-11, 13:05
Replies: 230
Views: 16,372
Posted By charybdis
It's not just a different polynomial, it's a...

It's not just a different polynomial, it's a completely different number (http://factordb.com/index.php?id=1100000000127408717) that was done last year...
Forum: Cunningham Tables 2021-09-09, 22:51
Replies: 76
Views: 30,763
Posted By charybdis
This didn't get anywhere, but I also ran 500k-10M...

This didn't get anywhere, but I also ran 500k-10M at P=5M, incr=420, which gave this:

n:...
Forum: Cunningham Tables 2021-09-08, 00:08
Replies: 76
Views: 30,763
Posted By charybdis
I used some top secret methods to figure out the...

I used some top secret methods to figure out the pre-sizeopt leading coefficients of Ed's polys, and he appears to have searched some range from 100M upwards - perhaps to 110M? - at incr=420 and a...
Forum: NFS@Home 2021-09-05, 13:15
Replies: 57
Views: 6,584
Posted By charybdis
Normally a hot spot is just a range with more...

Normally a hot spot is just a range with more special-q than expected. The variation in rels/q smooths out pretty well over a 5k range. I tried some more ranges and got the following:
Q Actual...
Forum: NFS@Home 2021-09-05, 11:21
Replies: 57
Views: 6,584
Posted By charybdis
Yield per special-q = 15438/323 = 47.796 ...

Yield per special-q = 15438/323 = 47.796
Expected number of special-q in 5k range at 20M = 5000/ln(20000000) = 297.42
Corrected yield estimate = 47.796*297.42 = 14215
Forum: NFS@Home 2021-09-05, 02:30
Replies: 57
Views: 6,584
Posted By charybdis
I'm not sure it's just 20M that's an anomaly...

I'm not sure it's just 20M that's an anomaly here. By complete coincidence, all four of the ranges that you picked contain more special-q than we would expect for a 5k range at those sizes; in total...
Forum: Aliquot Sequences 2021-09-04, 13:12
Replies: 524
Views: 63,545
Posted By charybdis
Inclined to agree, though if Oliver is hosting...

Inclined to agree, though if Oliver is hosting the final decision is up to him.
Forum: NFS@Home 2021-09-04, 01:34
Replies: 1,849
Views: 156,753
Posted By charybdis
3,748+ c204...

3,748+ c204 (https://mersenneforum.org/showthread.php?t=24548), the smallest unfactored Cunningham composite, doesn't seem to have had any attention in a while...
Forum: Aliquot Sequences 2021-09-04, 01:22
Replies: 524
Views: 63,545
Posted By charybdis
That's more like it: n:...

That's more like it:

n:...
Forum: Aliquot Sequences 2021-09-03, 20:22
Replies: 524
Views: 63,545
Posted By charybdis
Yes: technically on the algebraic side we're...

Yes: technically on the algebraic side we're looking at factorizations of (non-prime) ideals rather than of numbers. The product of two ideals is the set of all the finite sums of products of...
Forum: Aliquot Sequences 2021-09-03, 01:57
Replies: 524
Views: 63,545
Posted By charybdis
Best was 4.02e-15 which doesn't even beat the...

Best was 4.02e-15 which doesn't even beat the c202 record. I'll try the same range with P=10M; not expecting it to be better but I'd like to see how the performance compares.
Forum: NFS@Home 2021-09-03, 00:48
Replies: 1,849
Views: 156,753
Posted By charybdis
There are a bunch of 15e-suitable jobs in the...

There are a bunch of 15e-suitable jobs in the Cunningham base 2 extensions, but Batalov advised against (https://mersenneforum.org/showpost.php?p=571454&postcount=12) doing them before they get added...
Forum: Aliquot Sequences 2021-09-02, 18:58
Replies: 524
Views: 63,545
Posted By charybdis
An ideal is a subset of a ring that forms an...

An ideal is a subset of a ring that forms an additive subgroup and is closed under multiplication by arbitrary elements of the ring. But I don't think that's what you wanted :smile:

We have an...
Forum: Aliquot Sequences 2021-09-02, 17:55
Replies: 524
Views: 63,545
Posted By charybdis
This is a decent first approximation to what...

This is a decent first approximation to what required_excess does in practice, but strictly this isn't quite correct. It's not a proportion of total relations, but rather the number of relations...
Forum: Aliquot Sequences 2021-09-02, 13:28
Replies: 524
Views: 63,545
Posted By charybdis
I've done 2400 curves at 850M, no factor. I...

I've done 2400 curves at 850M, no factor.

I think the CADO default of P=10M for c200 is probably too big: Curtis's optimized c195 polyselect parameters have P=3M.
I'll do 0-5M with P=5M,...
Showing results 1 to 25 of 467

 
All times are UTC. The time now is 18:31.


Sat Sep 25 18:31:58 UTC 2021 up 64 days, 13 hrs, 0 users, load averages: 1.89, 1.53, 1.49

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.